Machine learning classification and FAQs retrieval for postoperative oral surgery questions

An artificial intelligence (AI)-based classification and retrieval system has the potential to map postoperative oral surgery questions to evidence-based answers to improve access to consistent information on postoperative conditions and management. This study was undertaken to show the development of an AI-based classification and retrieval system specifically designed to address inquiries related to postoperative oral surgery. The AI system was trained on a dataset encompassing 689 questions associated with various oral surgery procedures and employed machine learning algorithms such as Random Forest and Naïve Bayes models to accurately predict the corresponding operation for each query. Additionally, an augmentation technique was implemented to broaden the variety of oral surgery questions to train the model. The results indicated that the chatbot's performance achieved a moderate accuracy, ranging from 36.0% to 71.0% with the normal dataset. However, when trained with the augmented dataset, the chatbot's accuracy improved significantly, reaching between 88.0% and 97.0% in response to postoperative oral surgery questions. In conclusion, this AI-based classification and retrieval system demonstrates the potential to accurately classify postoperative oral surgery inquiries and, pending clinical validation, represents a beneficial future resource.